A lot of AI tools bring realistic images by following the description placed by the user. One of them is the Dall-e AI tool. The image AI tool counterpart from the same company that brought ChatGPT. Does Dall-e live up to its potential of bringing text prompts to actual graphics? Only one way to find out, and that is to test Dall-e.
I had a client that wanted to come up with a logo for a baseball firm. The client shared some concepts using Midjourney and had been very open about using AI tools to come up with a logo for them.
They wanted me to produce a badge-type baseball logo that would feature a rugged St. Bernard dog. These were the initial instructions. With that, I head over to Dall-e to create a baseball logo.
While Dall-e is marketed for its ability to generate realistic images, I wanted to try using the tool for a different application. Dall-e had some hit-and-miss when it came to producing the baseball logo. Let's get into the process.
Brainstorming a Logo Design
1. Possible Descriptions Used
My initial key descriptions revolved around their specific instructions. The first iterations were mainly to relay the concept. This helped me save 1-2 hours to generate the logo. I can go as far as creating a mascot as well for the baseball entity.
Initially, the descriptions used were "St. Bernard dog illustration biting baseball logo." I incorporated the verb "biting" to make the baseball look integrated into the dog. I wanted the dog and the baseball illustration to have some form of connection, so I used this prompt.
2. Interpretation of the AI Tool: First Impressions
Dall-e produced clean-looking graphics. The logo designs created via Dall-e were almost close to my initial idea. Some of the parts may need fine-tuning. Yet, it's already a presentable logo design.
The first one doesn't really fit the baseball logo I was looking for. A dog swallowing the bat doesn't provide relevant context.
As for the second design, it looks too basic for a baseball logo. It doesn't have the ruggedness or toughness I wanted to portray in the logo.
The badge-type logo was achieved in the third design iteration. However, I found the placement of the baseball elements a bit too misguided for this particular graphics.
Now, the last design generated by Dall-e had a sharp look to it. I could definitely use the styling and placement of elements, but not for this particular client who was looking for something in a badge form.
3. Refined Versions From the Initial Generated Images
I already had my eye on one of the initial concepts generated from Dall-e. From there, I used it as a base to come up with a refined version. The result from Dall-e showed uniformity among the graphics generated.
There were only the head of the dog and baseball elements distributed in the background. This was what I was trying to come up with. With the refined versions, it's now a matter of choosing which concept I'll move forward with for the logo.
Challenges Using the Tool and Its Impact on the Output
Dall-e is pretty much a straightforward AI tool that will let you generate images from a combination of word prompts. It will be a matter of choosing the right words to describe your concept, which I find a rather easy thing to do. However, in the test project I did, I had to come up with a logo design. This brings more dilemmas as I am not looking for an exact, word-for-word image. Rather, I want to create something subtle yet represents the baseball company I have to produce the logo.
While I attempted to look for baseball terms, the output generated was almost the same. It only has a few iterations, unless I change the descriptions used. However, the drastic changes are more likely to reflect another logo design concept. Some of these outputs were already far from the initial logo design concepts that I wanted to present to the client.
Branding Components of the Logo
Can It Fully Create Miniscule Details?
This is where Dall-e can be difficult to play around. I wanted some minor iterations for the logo design. Any simple details will be elaborately interpreted, bringing drastic changes to the logo. The logo for the baseball entity had to be fine-tuned to include minor details. This precise level of editing can be difficult, if not at all impossible, with just a few text prompts. I'm left with no option other than to manually edit or remove any details from the logo created via Dall-e.
Brand Essence From the AI-Generated Logo
Another challenge is when it comes to interpreting the brand's essence. I cannot fully trust Dall-e to generate a logo based on vague statements. The images generated may either be too direct to the point or do not meet the look and feel I'm targeting. There is a lack of essence coming from the logo generated from Dall-e. The haphazardly done image is very obvious coming from an AI tool - it looks awkward, and you cannot define any substance from the graphics.
Reflecting the Brand’s Identity
A company's brand identity can't simply be reflected by using the exact words. Dall-e will interpret them as is, according to the choice of words used. Any subtle hints of creative play or strength that I wanted to incorporate in the baseball logo may not be accurately depicted by Dall-e. I cannot interpret the brand identity, values, or target audience via descriptive text. The AI tool lacks the creative instinct to incorporate branding elements in the image.
Importance of Human Intervention
Dall-e is indeed a useful tool to fast-track the creative aspect of conceptualizing the logo. I cannot stress this enough: the logo will still need to be refined and polished more. Thus, there's a need to have human intervention. I still need to put my creative juices out there and incorporate them into the logo design to make it more defined and coherent. The expertise of a graphic designer is still needed to fine-tune the output generated via Dall-e.
I can use Dall-e to generate initial design concepts. This would make the job faster, and I can get feedback from the client right away if it's the look they wanted for their logo. This back-and-forth of logo design pegs is crucial. In a way, I can understand their ideas better. And, with the Dall-e tool, I can generate concepts right off the bat.
Still, the Dall-e AI tool is far from being a 100% reliable machine to produce logos and images. It can generate designs based on the descriptions. However, some of these descriptions could have overlapping meanings. At times, Dall-e misinterprets them. Especially with the baseball logo that I had to create. I had to also make sure that the values and branding of the baseball firm were present in the logo.
Having the creative intuition of a human graphic artist is still crucial in the logo design process. I still had to tweak the images to make it fitting to use as a logo. While Dall-e helped me fast-track the initial concept stages, the final logo output still needs my ideas. The Dall-e AI tool could be useful for client communication and presentation. A visual representation of ideas is a far better method than text descriptions, which Dall-e was able to help me with the process.
But then again, the finishing touches and overall concept still need human expertise. The Dall-e AI tool cannot comprehend vague concepts like branding and interpret them into a baseball logo. Still, the collaboration between using the Dall-e AI tool and the final creative decision from a human graphic artist is a method that we'll be seeing more in the future.
Bhavya Shah is a graphic designer focused on brand strategy and fascinated by UX research. She thrives on combining research, data, and design to create impactful visual solutions.
